Issue link: https://iconnect007.uberflip.com/i/1511130
12 DESIGN007 MAGAZINE I NOVEMBER 2023 normal for tolerance, but they missed the part of the spec that says 10% or 1 mil, whichever is greater. Fabricators can't hold that tight a layer-to-layer tolerance when it's below a cer- tain layer thickness. Below about a 10-mil thickness, the best fabricators can do layer-to-layer is 1 mil for nominal processing. If you want to hold a tighter tolerance, you're paying for 100 to get five boards. at's just a couple of exam- ples. Another is overly tight hole tolerances: "I want to have 150-mil diameter hole plus or minus one mil." Again, it's unreasonable, right? At several companies where I've worked or consulted, they still have mechanical engineers do the documentation for the boards. So, the board designer designs the board and passes it over to let the ME do the documentation and follow their s t a n d a r d d o c u - mentation template. But the designers don't real- ize that a machine shop that's milling a chunk of aluminum can absolutely hold plus or minus 1 or 2 mils of tolerance. ere's not a lot of understanding of what the manufacturing capabilities are. ey tried to hold the manufacturing nuances that they're used to in mechanical engineering, apply those to board design and board manufacturing, and find themselves overly constrained. Shaughnessy: How did the MEs end up in this position? Moyer: If you go back to the early days of CAD tools, the MCAD tools were set up from day one to create good documentation packages. e ECAD tools would output a document, but it didn't have good documentation capabilities like GDT control frames. In early ECAD tools, if you had 3D at all, it was simple extruded rect- angles. "Here, put a box on your board where your part is extruded up, so you know what the thickness is." at represented your part, whereas the MEs had full 3D modeling to build very complex, realistic structures, so they can do all of that. A lot of companies just said, "You know what? Just send the board to the ME." ey already had a template defined and the ME could do all of these little detailed views and so on. Shaughnessy: What's the solution here? Moyer: e solution is to learn the standards and learn the manufacturing processes. Take all the IPC training classes that you can to learn these processing capabilities and where the limitations are. en the board designer can have intelligent, knowledgeable discussions with the other stakeholders—the EEs, MEs, managers, and so on. You have to be able to say, "I understand where you're coming from. However, this is the manufacturing limita- tion. If you spec it this way, it's a big cost-adder to the board. But if we docu- ment it this way, we w i l l g e t t h e s a m e effect that we want, but now we've simplified our board design and constrained it just enough for what we need without over- constraining." Happy Holden: Andy, I think what's complicat- ing the idea of simplifying your design is that BGAs and other part miniaturization and the shrinking pitch of connections all sneak up on designers. A lot of times, they design the whole board and get everything connected except for the last 5%. en they run into via starvation, and they're forced to go with laser- drilled blind vias. I can always tell that, if it's a complex board with 127 laser-drilled blind vias, that's a clear indication that the designer never intended this to be an HDI board. He didn't know about the simple models they can use that would have told him ahead of time that the size of the board, the pitch that he's work- The solution is to learn the standards and learn the manufacturing processes.